Improved Microstructure and Properties of 6061 Aluminum
Alloy Weldments Using a Double-Sided Arc Welding
Process
Y.M. ZHANG, C. PAN, and A.T. MALE
Due to its popularity and high crack sensitivity, 6061 aluminum alloy was selected as a test material
for the newly developed double-sided arc welding (DSAW) process. The microstructure, crack sensitivity, and porosity of DSAW weldments were studied systematically. The percentage of fine equiaxed
grains in the fully penetrated welds is greatly increased. Residual stresses are reduced. Porosity in
the welds is reduced and individual pores are smaller. It was also found that the shape and size of
porosity is related to solidification substructure. In particular, a weld metal zone with equiaxed
grains tends to form small and dispersed porosity, whereas elongated porosity tends to occur in
columnar grains.
I. INTRODUCTION
AS one of the most commonly used heat-treatable aluminum alloys, 6061 is available in a wide range of structural
shapes, as well as sheet and plate products. Typically, it is
used in autobody sheet, structural members, architectural
panels, piping, marine applications, screw machine stock,
and many other applications.[1] Generally, this alloy is easily
welded by conventional arc welding processes (gas metal
arc welding and gas tungsten arc welding (GTAW)) and highenergy processes (laser-beam and electron-beam welding).
However, certain characteristics, such as solidification cracking, porosity, heat-affected zone (HAZ) degradation, etc.,
must be considered during welding, due to the greater
amount of alloying additions used in this alloy.[2–5] Because
of high-energy density and low overall heat input, laser beam
and electron beam welding processes possess the advantage
of minimizing the fusing zone and HAZ[5] and producing
much deeper penetration than arc welding processes.[6] However, their high cost limits their usage in industry.
Currently, the authors have developed a new welding
process called “double-sided arc welding” (DSAW).[7–9] In
this process, as shown in Figure 1, two torches (such as
plasma arc torch and gas tungsten arc (GTA) torch, or dual
GTA torches) are placed on the opposite sides of a base metal
plate to increase penetration. They are directly connected to
two terminals of a single power supply. The welding current
loop becomes negative terminal - anode torch - workpiece cathode torch - positive terminal instead of the conventional
negative terminal - anode torch - cathode workpiece - positive terminal. As a result, current flow concentrates the arc
and improves weld penetration, resulting in a reduction in
heat input. For example, in order to penetrate 6.5-mm-thick
Al plate, regular AC GTAW needs two passes, but AC double - sided GTAW requires only one pass.[8] In addition, the
Y.M. ZHANG, Associate Professor of Electrical Engineering, C. PAN,
Research Associate, and A.T. MALE, Professor and Director, are with the
Center for Robotics and Manufacturing Systems, University of Kentucky,
Lexington, KY 40506-0108. C. PAN is on leave from Wuhan University
of Technology, P.R. China.
Manuscript submitted October 21, 1999.
METALLURGICAL AND MATERIALS TRANSACTIONS A
heat input in each pass for regular AC GTAW is approximately twice the heat input needed by the later.[8] The heat
input is reduced to 25 percent. This process may provide a
method to weld aluminum alloys without filler metal addition and to generate positive effects on productivity, cost, and
weld quality. Extensive experiments have been performed on
different metals and alloys using the DSAW process. Some
unique characteristics and advantages have been obtained.
For example,[7] on 6.4-mm-thick aluminum plates, the
DSAW achieves 5.2-mm depth with 6-mm width, while
regular variable polarity plasma arc welding (VPPAW) penetrates 3 mm with 8-mm width. The depth-to-width ratio is
nearly doubled.
In the present work, the microstructures, solidification
behavior, and cracking sensitivity of the 6061 aluminum
alloy welded joints were studied systematically by comparison between normal arc welding process and the present
DSAW process.
II. EXPERIMENTAL PROCEDURE
The 6061 aluminum alloy studied in the present experiment was the commercial plate 6061-T651 (wt pct: 0.28Cu,
0.6Si, 1.0Mg, 0.20Cr, and bal Al) in thickness of 4.76, 6.4,
and 9.5 mm.
The DSAW with dual GTA torches was performed without
filler metal addition. The two terminals of the square wave
constant current AC power supply were connected to two
regular GTA torches. The polarity ratio was 15 to 15 ms.
The maximum current of the power supply was 150 A at
arc voltage of 50 V. Uphill (vertical) welds were made in
the butt joints from both sides with a shielding gas of pure
argon at the flow rate of 12 L/min. In addition to DSAW,
VPPAW was also used to make comparative welds with
1.5 L/min plasma gas flow rate and a 2.57-mm (0.093-in.)
diameter orifice. Table I lists major welding parameters and
conditions for both DSAW and VPPAW. Because of the
difference in penetration capability and process characteristics, different parameters and conditions were used for
DSAW and VPPAW.
Samples were mechanically polished and electrolytically
etched with a solution of 20 pct hydrofluoric acid 1 80 pct
VOLUME 31A, OCTOBER 2000—2537
Fig. 1—The principle of double-sided GTA welding process.
water. The microstructures, fracture surfaces, and porosity
were examined using a Nikon Epiphot 300 optical metallurgical microscope and a Hitachi S-3200 scanning electron
microscope (SEM), operated at 20 kV.
III. RESULTS AND DISCUSSION
A. Solidification Behavior
Generally, solidification behavior in the weld pool is determined by several factors, such as thermal gradient in the
liquid, GL , solidification growth rate, R, chemical composition, pool shape, etc.[10,11] However, all of these depend upon
the welding process. Different processes produce different
solidification structures, thus different mechanical properties
of joints. Among existing arc welding processes for aluminum, the VPPAW welding process achieves the deepest penetration. Hence, it was selected for comparative studies with
the unique solidification characteristics of DSAW.
Figure 2 illustrates the solidification structures around
the fusion boundary with different welding processes and
different welding conditions. In the bead-on-plate VPPAW
joint and partially penetrated double-sided GTAW joint, the
microstructures of the welds consisted of well-developed
cast columnar structures, which nucleated and grew epitaxially from the solid-liquid boundary or partially melted grains
toward the upper surface, as shown in Figures 2(a) and
(b). Observation indicated that the columnar grains typically
comprised over 80 pct of the whole weld metal zone. Equiaxed grains only formed in a small area around the center.
In addition, the size of the columnar grains increased toward
the fusion boundary. This is the result of epitaxial solidification, which grows toward the center in a direction along the
maximum thermal gradient.[10,11] The growth rate increases
from zero at the fusion boundary to a maximum value at
the weld center.[11]
As observed in Figure 2(c), in the fully penetrated
DSGTAW joint, only a very narrow columnar structure exists
2538—VOLUME 31A, OCTOBER 2000
Fig. 2—Microstructures around the fusion boundary: (a) sample S-5,
VPPAW; (b) sample S-3, DSGTAW, partial penetration; and (c) sample S1, DSGTAW, full penetration.
along the fusion boundary. In most of the weld metal zone
(typically over 70 pct), fine equiaxed grains become the
major solidification structure. Figure 3 shows a typical morphology of the fine equiaxed grain in the center of the weld
metal zone.
Generally, fine equiaxed grains tend to reduce solidification cracking and improve mechanical properties of the
welded joint. This is due to the fact that, in fine-grained
materials, low melting point segregates tend to be distributed
over a larger grain boundary area, and equiaxed grains
accommodate strains more uniformly or permit easier transport of liquid between grains.[12] However, unlike in casting,
the natural occurrence of columnar-to-equiaxed transition in
METALLURGICAL AND MATERIALS TRANSACTIONS A
Table I. Welding Conditions
Sample
Welding
Method
Thickness
(mm)
Arc Voltage
(V)
Welding
Current (A)
Welding
Speed (mm/s)
Joint Type
Penetration
S-1
S-2
S-3
S-4
S-5
DSGTAW
DSGTAW
DSGTAW
DSGTAW
VPPAW
6.4
6.4
6.4
9.5
4.76
47
47
47
48
36
145
145
145
150
EP 5 100
EN 5 80
4.2
6
7.5
2
4.7
butt
butt
butt
butt
bead-on-plate
full
full
partial
full
full
Fig. 3—Morphology of equiaxed grains in the weld metal zone of sample
S-1.
the grain structure of weld is not very common.[10,13] It is
well known that the Ti and Zr[14–18] and Cu, Cr, and Mn[19,20]
alloying additions have an effect on grain refinement in Albase casting and welding. Pearce and Kerr[21] used magnetic
stir to increase the fraction of fine equiaxed grain in aluminum alloys. Clark et al.’s[22] research demonstrated that a
columnar-to-equiaxed transition is favored in GTA welding
in Al-Cu alloy by using high current and welding speed
combination, increasing copper content, and increasing the
weight percent of the nucleating agent for equiaxed grains.
Brooks[11] found that a large equiaxed zone existed in the
6061 Al weld because of a high degree of constitutional
supercooling.
For a given alloy system, the morphology of the solidification structure is controlled by the solidification parameters—
the solidification growth rate R and the thermal gradient in
the liquid GL. That is to say, the ratio of the two parameters
GL /R changes from a maximum value at the fusion boundary
to a minimum along the center of the weld. These changing
solidification conditions result in a weld solidification structure changing from planar at the weld boundary to columnar
dendrite and then to equiaxed dendrite grain along the
weld center.[13]
For the present DSGTAW process, extensive experimental
work has revealed that the fraction and width of the fine
equiaxed grain region also gradually increased in the weld
metal zone along with the increase in depth of penetration.
It is known that when the penetration increases, the amount
of the melted metal increases. Such an increase in the amount
of the melted metal helps heat the workpiece before cooling.
Hence, the thermal gradient during cooling is reduced. This
tends to increase the amount of fine equiaxed grains. However, equiaxed grains are observed throughout nearly the
METALLURGICAL AND MATERIALS TRANSACTIONS A
whole weld metal zone. This cannot be explained by the
reduced thermal gradient alone. The authors believe that the
alternative fluid flow in the weld pool may be the major
cause of such a formation of the equiaxed grain zone. In
fact, in conventional arc welding, the welding current is
largely grounded through the surface of the workpiece and
little current flows through the depth of the weld pool. In
the DSAW process, the welding current must directly flow
through the weld pool from one side to the other side of the
workpiece. The presence of the welding current inside the
weld pool must cause an electromagnetic force driven fluid
flow in the weld pool. Due to the varying polarity of the
current, the direction of such fluid flow must change periodically. Such change may tend to generate a stirring effect in
the weld pool.[8] Then the nucleation and growth of the
grains during solidification becomes isotropic and forms the
fine equiaxed grains in the greater part of the weld metal
zone. This solidification characteristic will benefit the properties of the 6061 aluminum alloy welded joint, as will be
discussed in Section B.
B. Solidification Cracking Sensitivity
The popularity and higher solidification cracking sensitivity of 6061 aluminum alloy weldments are other factors that
attract many researchers.[2,3,5,8,23] Generally, solidification
cracking occurs when higher levels of thermal stress and
solidification shrinkage are present during welding.[1] It is
influenced by a combination of mechanical, thermal, and
metallurgical factors. In practice, the solidification cracking
sensitivity of aluminum alloy weldments is determined by
the chemical composition and weld conditions. For 6061
alloy, the greater amount of alloying additions of Mg and
Si increases its cracking sensitivity. The primary methods
for eliminating cracking in aluminum welds are to control
weld metal composition through filler alloy additions and
to use low heat input by using a special welding process,
such as electron-beam or laser-beam welding.[1,3,5] In practice, some other methods, such as arc oscillations,[24–26] electromagnetic stirring,[21,27] external local heating,[28] and
mechanical vibration,[29] are also used.
Figures 4 and 5 illustrate the weld surface and crosssectional appearance of welds made using different processes. It is clear that the DSAW process has the higher
cracking resistance. The cracking in the bead-on-plate
VPPAW joint is typical of solidification cracking, which
appears along the center of the weld metal zone. As discussed
previously, the cracking is mainly produced by two factors,
stress conditions and metallurgical factors.
In general, the stress concentration in the welded joint of
aluminum alloy is induced in two ways: thermal stress,
VOLUME 31A, OCTOBER 2000—2539
(a)
(b)
Fig. 4—Morphologies of the weld surfaces: (a) sample S-5, VPPAW; and
(b) sample S-1, DSGTAW.
Fig. 5—Morphologies of the cross section of the welds: (a) sample S-5,
VPPAW; and (b) sample S-1, DSGTAW.
due to the high coefficient of thermal expansion and large
solidification shrinkage, almost twice that of steel. When an
aluminum alloy plate is welded using a normal arc welding
process, the molten pool typically is V-shaped, as shown in
Figure 5(a). Shrinkage forces within the V-shaped zone cause
2540—VOLUME 31A, OCTOBER 2000
Fig. 6—Sketch map of the shrinkage force in the welds: (a) VPPAW, beadon-plate; and (b) DSGTAW, butt, no filler.
the plate to have an angular distortion. The shrinkage induced
stresses increase from bottom to top surface,[30,31] as shown
in Figure 6(a). If the plate is constrained during welding,
the distortion will decrease; however, the residual stress in
the weld zone will greatly increase.[30] However, in the case
of the DSAW process, two GTAW torches act upon the
aluminum plate simultaneously and symmetrically. This
means that shrinkage forces are symmetrical in the weld zone
during cooling, as illustrated by Figure 6(b). This unique
phenomenon associated with DSAW minimizes the transverse distortion and the residual stress in the weld pool. It
helps reduce cracking sensitivity.
The solidification microstructure is another critical factor
influencing cracking sensitivity in aluminum alloy weldments.[2,3,10,11] The DSAW process produces fine equiaxed
grains in the weld metal zone, as shown in Figures 2 and
3. This microstructure is known to improve solidification
cracking resistance. Observation of fracture surfaces
VPPAW welds revealed that the solidification cracking
nucleated, propagated, and disbonded along the columnar
grain boundaries, as shown in Figure 7(a). Also, under higher
magnification, as shown in Figure 7(b), secondary cracks
and some secondary eutectic phases were observed. The
secondary phase eutectic constituents, such as Mg2Si and Si,
surround the columnar structure and constitute a significant
fraction of the part surface. This implies that solidification
cracks initiated at a time very close to or after final
solidification.[2]
For DSAW, the desired dimples, which indicate plastic
deformation and higher toughness, are observed on the fracture surface of the weld metal zone. The fracture surface
also appears to have less porosity with smaller pores, as
shown in Figure 8. Both improvements are attributed to the
fine equiaxed grains exhibited in the DSAW weldments.
C. Porosity in the Weld Metal Zone
It is desirable to limit porosity defects in aluminum weldments.[1,3–5,32–35] Porosity forms when hydrogen gas is
METALLURGICAL AND MATERIALS TRANSACTIONS A
Fig. 7—SEM micrographs of the fracture surface in the weld metal zone
of sample S-5 (VPPAW, bead-on-plate): (a) low magnification and (b)
high magnification.
Fig. 8—SEM micrograph of the fracture surface in the weld metal zone
of sample S-1 (DSPAW, full penetration).
METALLURGICAL AND MATERIALS TRANSACTIONS A
Fig. 9—Low magnification of porosity in the weld metal zones: (a) sample
S-5, VPPAW, bead-on-plate; and (b) sample S-1, DSGTAW, full penetration.
entrapped during solidification.[1] Hydrogen is absorbed into
the molten pool during welding because it is highly soluble
in molten aluminum. Gas pores form during solidification,
because solubility in the solid is less than in the melt and
hydrogen is rejected from the solid to the melt causing
localized supersaturation, bubble nucleation, and growth.
Increase in porosity is generally associated with high
humidity and poor surface preparation. Use of inert gases
to shield the weld pool can reduce porosity. In the present
study, no special attention was paid to surface cleaning and
shielding gas. The conditions were unchanged for VPPAW
and DSAW. Bulk pores were not found in the weld metal
zones of DSAW weldments. However, from Figure 9, it can
be observed that the pore size in the plasma arc weld zone
is significantly larger than that found in DSAW joints, that
is, about 35 mm in the weld of VPPAW and 10 mm in the
weld of DSAW. Higher magnification using an SEM shows
clearly that columnar grains tend to produce an elongated
porosity, whereas the equiaxed grains tend to form a smaller
and more dispersed porosity, as shown in Figure 10.
According to the theory of formation of gas porosity in
aluminum alloys,[36–38] the long pores precipitate at a later
stage of solidification, when crystals/dendrites are growing
throughout the melt and are influenced by the hydrogen
VOLUME 31A, OCTOBER 2000—2541
IV. CONCLUSIONS
Compared to bead-on-plate VPPAW welding of 6061 aluminum alloy, the DSAW process has the following
advantages.
1. The percent of equiaxed grains is increased and columnarto-equiaxed grain transition occurs earlier than in partially
penetrated DSAW and VPPAW welds.
2. Hot cracking sensitivity is reduced by minimizing residual
stresses in the weld, as a result of the symmetrical temperature profile produced during double-sided welding.
3. Pores are smaller and more dispersed among the equiaxed
dendrites produced by DSAW with full penetration than
in the partially penetrated DSAW and VPPAW welds.
ACKNOWLEDGMENTS
This work is supported by the National Science Foundation (Grant No. DMI 9812981) and the Center for Robotics
and Manufacturing Systems (CRMS) at the University of
Kentucky. The authors express appreciation to Dr. S.B.
Zhang for his cooperation in this work.
REFERENCES
Fig. 10—SEM micrographs of porosity in the weld metal zones: (a) sample
S-5, VPPAW; and (b) sample S-1, DSGTAW, full penetration.
enrichment and the shrinkage pressure in the columnar interdendritic area during solidification. On the other hand, the
small and fissured pores precipitate at a very late stage of
solidification, the bubble growth is severely limited, and the
shape is determined by the interdendritic space available.
Therefore, the large amounts of equiaxed grains solidified
in the weldments of the DSAW process tend to form the
small and dispersed porosity.
In addition, the characteristics of the DSAW process, such
as its special bidirectional buoyancy, alternating electromagnetic force inside the weld pool, and dual surface tensions,
may also control the formation of porosity. The hydrogen
gas can escape from both melted sides of the weld pool, and
the amount of pores are reduced. Hence, the solidification
behavior of the DSAW process and the resultant size and
distribution of porosity in the weld metal zone differs from
other welding processes.
Comparing to different welding processes and welding
conditions, the fully penetrated joint welded with the DSAW
process, which exhibits a high proportion of equiaxed grains
as discussed previously, contains smaller pores and less total
porosity. This characteristic helps improve the mechanical
properties of the joint.
2542—VOLUME 31A, OCTOBER 2000
1. ASM Specialty Handbook: Aluminum and Aluminum Alloys, J.R.
Davis, ed., ASM INTERNATIONAL, Materials Park, OH, 1994, pp.
376-419.
2. J.A. Brooks and J.J. Dike: Proc. 5th Int. Conf. on International Trends
in Welding Research, Pine Mountain, GA, June 1998, ASM INTERNATIONAL, Materials Park, OH, 1999, pp. 695-99.
3. D.M. Douglass, J. Mazumder, and K. Nagarathnam: Proc. 4th Int.
Conf. on Trends in Welding Research, Gatlinburg, TN, June 1995,
ASM INTERNATIONAL, Materials Park, OH, 1996, pp. 467-78.
4. J.S. Kim, T. Watanabe, and Y. Yoshida: J. Mater. Sci. Lett., 1995, vol.
14, pp. 1624-26.
5. A. Hirose, H. Todaka, and K.F. Kobayashi: Metall. Mater. Trans. A,
1997, vol. 28A, pp. 2657-62.
6. Welding Handbook, 8th ed., R.L. O’Brien, ed., American Welding
Society, Miami, FL, 1995, vol. 2, pp. 671-710.
7. Y.M. Zhang and S.B. Zhang: Weld. J., 1998, vol. 77, pp. 57-61.
8. Y.M. Zhang and S.B. Zhang: Weld. J., 1999, vol. 78, pp. 202s-206s.
9. Y.M. Zhang and S.B. Zhang: Proc. 5th Int. Conf. on International
Trends in Welding Research, Pine Mountain, GA, June 1998, ASM
INTERNATIONAL, Materials Park, OH, 1999, pp. 271-75.
10. S.A. David and J.M. Vitek: Proc. 3rd Int. Conf. on International Trends
in Welding Science and Technology, Gatlinburg, TN, June 1992, ASM
INTERNATIONAL, Materials Park, OH, 1993, pp. 147-56.
11. J.A. Brooks: Proc. 4th Int. Conf. on Trends in Welding Research,
Gatlinburg, TN, June 1995, ASM INTERNATIONAL, Materials Park,
OH, 1996, pp. 123-34.
12. S. Kou: Welding Metallurgy, John Wiley & Sons, New York, NY,
1987, p. 211.
13. S.A. David and J.M. Vitek: Int. Mater. Rev., 1989, vol. 34, pp. 213-45.
14. G.W. Delamore and R.W. Smith: Metall. Trans., 1971, vol. 2, pp.
1733-38.
15. J. Cisse, H.W. Kerr, and G.F. Bolling: Metall. Trans., 1974, vol. 5,
pp. 633-41.
16. T. Ganaha, B.P. Pearce, and H.W. Kerr: Metall. Trans. A, 1980, vol.
11A, pp. 1351-59.
17. H. Yunjia, R.H. Frost, D.L. Olson, and G.R. Edwards: Weld. J., 1989,
vol. 68, pp. 280s-289s.
18. M.J. Dvornak, R.H. Frost, and D.L. Olson: Weld. J., 1989, vol. 68,
pp. 327s-335s.
19. H.T. Kim, S.W. Nam, and S.H. Hwang: J. Mater. Sci., 1996, vol. 31,
pp. 2859-64.
20. H.T. Kim and S.W. Nam: Scripta Mater, 1996, vol. 34, pp. 1139-45.
METALLURGICAL AND MATERIALS TRANSACTIONS A
21. B.P. Pearce and H.W. Kerr: Metall. Trans. B, 1981, vol. 12B, pp.
479-89.
22. J. Clarke, D.C. Weckman, and H.W. Kerr: Proc. 5th Int. Conf.
on International Trends in Welding Research, Pine Mountain, GA,
June 1998, ASM INTERNATIONAL, Materials Park, OH, 1999, pp.
72-76.
23. W.D. Fei and S.B. Kang: J. Mater. Sci. Lett., 1995, vol. 14, pp. 1795-97.
24. S. Kou and Y. Le: Weld. J., 1985, vol. 64, pp. 51-55.
25. S. Kou and Y. Le: Metall. Trans. A, 1985, vol. 16A, pp. 1345-52.
26. S. Kou and Y. Le: Metall. Trans. A 1985, vol. 16A, pp. 1887-96.
27. T.N. Tayalajan and C.E. Jackson: Weld. J., 1972, vol. 51, pp. 337s-340s.
28. I.E. Hernandez and T.H. North: Weld. J., 1984, vol. 63, pp. 84s-90s.
29. D.C. Brown: Weld. J., 1962, vol. 41, pp. 241s-250s.
30. J. Puchaicela: Weld. J., 1998, vol. 77, pp. 49-52.
METALLURGICAL AND MATERIALS TRANSACTIONS A
31. S.C. Gambrell and K. Kavikondala: Weld. J., 1996, vol. 75, pp.
109s - 114s.
32. R.F. Ashton, R.P. Wesley, and C.R. Dixon: Weld. J., 1975, vol. 54,
pp. 95s-98s.
33. K. Norenberg and J. Ruge: Aluminum, 1992, vol. 68, pp. 406-10.
34. P.A. Molian and T.S. Srivatsan: Aluminum, 1990, vol. 66, pp. 69-71.
35. M. Pastor, H. Zhao, R.P. Martukanitz, and T. Debroy: Weld. J., 1999.
vol. 78, pp. 207s-216s.
36. N. Roy, A.M. Samuel, and F.H. Samuel: Metall. Mater. Trans. A,
1996, vol. 27A, pp. 415-29.
37. K. Tynelius, J.F. Major, and D. Apelian: Trans. Am. Foundrymen’s
Soc., 1993, vol. 101, pp. 401-13.
38. X.G. Chen and S. Engler: Trans. Am. Foundrymen’s Soc., 1994, vol.
102, pp. 673-82.
VOLUME 31A, OCTOBER 2000—2543